diff --git a/backend/services/booking_scraper.py b/backend/services/booking_scraper.py
index b314056..ad8d3dd 100644
--- a/backend/services/booking_scraper.py
+++ b/backend/services/booking_scraper.py
@@ -524,15 +524,20 @@ def _is_hotel_page_mode(db: Session) -> bool:
     return (row and row.config_value) == 'playwright_hotel_page'
 
 
-def _effective_concurrency(db: Session, n_jobs: int) -> int:
-    """Clamp configured concurrency to the workload, and force serial when the
-    proxy is off — N workers would share one IP and hammer it, worse than 1."""
+def _effective_concurrency(db: Session, n_jobs: int, require_proxy: bool = True) -> int:
+    """Clamp configured concurrency to the workload.
+
+    require_proxy=True (search-results mode): force serial without a proxy because
+    all workers would hit the same search URL from the same IP.
+    require_proxy=False (hotel-page mode): allow parallelism even without proxy —
+    each worker hits a different hotel's URL so concurrent requests look like normal
+    multi-tab browsing rather than a hammered aggregation endpoint.
+    """
     configured = get_scraper_concurrency(db)
     if configured <= 1 or n_jobs <= 1:
         return 1
-    # get_scraper_backend does no I/O beyond the config read; safe to probe.
-    if not get_scraper_backend(db)._proxy_enabled():
-        logger.info("Proxy disabled — running scrape serially (parallelism needs per-worker IPs)")
+    if require_proxy and not get_scraper_backend(db)._proxy_enabled():
+        logger.info("Proxy disabled — running search-results scrape serially (parallelism needs per-worker IPs)")
         return 1
     return max(1, min(configured, n_jobs))
 
@@ -767,7 +772,7 @@ async def _run_hotel_page_scrape(
     )
     db.commit()
 
-    concurrency = _effective_concurrency(db, len(hotels))
+    concurrency = _effective_concurrency(db, len(hotels), require_proxy=False)
     logger.info(
         f"Hotel-page scrape {from_date}..{to_date}: "
         f"{len(hotels)} hotels × {len(dates)} dates = {len(hotel_date_jobs)} jobs, "